Copy Link
Add to Bookmark
Report
TCP-IP Digest Vol. 1 No. 19
TCP/IP Digest Monday, 19 Apr 1982 Volume 1 : Issue 19
Today's Topics:
InterNet Protocol Transition Workbook Availible from NIC
TOPS-10 Implementation of TCP/IP Slated by U.S. Air Force
Misinformation corrected: 3-Com, Navy Plans, SRI Plans
Further comments on DTI's ACCESS Offering
Comments on Service Specifications
TCP/IP from Scratch -- Request for Help
----------------------------------------------------------------------
LIMITED DISTRIBUTION
For Research Use Only --- Not for Public Distribution
----------------------------------------------------------------------
From: POSTEL at USC-ISIF
Subject: Internet Protocol Transition Workbook
To: tcp-ip at BRL
A book containing just about all of the ARPA Internet protocols has
been put together by the Network Information Center. This book includes
IP, TCP, Telnet, FTP, Mail, UDP, TFTP, and Name Server Protocols. It
also includes information about host tables, assigned numbers, and other
reference information. The book can be obtained from the Network Information
Center by sending a message with your name and address to NIC@NIC.
--jon.
------------------------------
From: PROVAN at WPAFB-AFWAL
Subject: tops-10 implementation slated
To: tcp-ip at BRL
The air force has allocated me to implement ip/tcp for
tops-10. I'm hoping to get it up before january 1. interested
parties should get in touch with me.
------------------------------
Sender: CERF at USC-ISI
Subject: Re: TCP-IP Digest, Vol 1 #18
From: CERF at USC-ISI
To: TCP-IP at BRL
IT IS MY UNDERSTANDING THAT 3COM INTENDS TO TRACK ANY CHANGES IN
THE DOD STANDARD PROTOCOLS INCLUDING TCP AND IP.
VINT CERF DARPA/IPTO
------------------------------
From: Mike Muuss <Mike@brl>
To: Vint <Cerf@usc-isi>
cc: TCP-IP at Brl
Subject: 3Com tracking DoD Standards
I stand corrected. That is a welcome message indeed. Thanks!
Best,
-Mike
------------------------------
From: ron at NOSC-CC (Ronald L. Broersma)
Subject: NAVY no longer attempting plan #1
To: tcp-ip at brl
Mike,
In the recent DIGEST, you said that NAVY and SRI were doing #1 of the
three efforts for TCP/IP on Unix V7. The NAVY has just decided to
buy 7 VAX 11/750s to replace most of the PDP 11/40s and go with the
Berkeley TCP/IP software (4.2BSD) whenever that is released.
--Ron
------------------------------
From: croft at SRI-TSC
To: tcp-ip at brl, croft at SRI-TSC
Subject: Re: TCP-IP Digest, Vol 1 #18
Mike,
In your latest TCP-IP digest you mention that SRI is going to be using
the BBN user-mode TCP. This is incorrect. What we are doing is
converting the Berkeley 4.2 VAX TCP/IP to run in an overlayed
2.81 BSD PDP-11 environment. As a stop-gap we have one UNET host
running TCP/NCP (SRI-PRMH). In about 2 months we hope to have the
Berkeley TCP running on all our VAX's and 11's.
--Bill Croft
[ Looks like lots of folks have updated their plans since my last contact
with them... Sorry to have distributed out of date information. -Mike ]
------------------------------
Subject: TCP/IP for VAX/VMS
From: BOLTE at OFFICE-8
To: TCP-IP at BRL
I recently received (indirectly) a copy of DTI's ACCESS ARPANET
Software Products paper. It was an unsolicited response to
someone's plans for a VAX 11/780. It seems that DTI reads the
Commerce Business Daily.
In addition to the comments that Gary Grossman made in the last
TCP-IP Digest, here are some more:
Documentation: *ACCESS Site Administrator's Giude
& *ACCESS User's Guide
Training: They expect to offer ACCESS-T training course by this month.
Pricing: ACCESS-N (NCP version) & ACCESS-T (TCP/IP version)
each cost $15,000.
Upgrade from ACCESS-N to ACCESS-T is $6,000.
ACC LH/DH-11 hardware (Assoc.Comp.Cons.) is $6,500.
Additional ACCESS-T's at the same site cost $6,000. each.
Software Support: ACCESS-T: $4000/yr or $400/mo
Above prices quoted as of Jan '82.
An additional POC is: Gary Tauss (217) 384-8500.
Digital Technology Incorporated
302 E. John St.
Champaign, Il 61820
...Bill
------------------------------
From: Walt <Haas at UTAH-20>
Subject: Re: Service Specifications
To: ihnss!npois!srk at UCB-C70
cc: TCP-IP at BRL
The major objection to the three-arrow approach proposed by the HILI
committee is precisely that the response generated by the server layer
does not in fact carry any information to indicate whether the
recipient of the connection has accepted it. For example, my X.25
implementation for the DEC-20 verifies that the original user can in
fact connect to the system before it returns the "CONNECT.response"
primitive to the network; hence the "CONNECT.confirm" primitive
received by the original user layer indicates that there has in fact
been some response from the remote user layer. This is my
understanding of how the ISO reference model is intended to work.
The three arrow model which the HILI committee proposes strips the
"CONNECT.confirm" primitive of most of its useful information content,
and so renders it vitually useless. In the HILI committee's proposed
model, this primitive indicates only that the remote user layer is
capable of absorbing "CONNECT.indication". The original user receives
absolutely no useful information about whether the remote user has,
for example, enough resources to actually establish a connection. In
fact this "CONNECT.confirm" is equivalent to a server-level
acknowledgement, and as such is scarcely worth the trouble of passing
to the original user.
------------------------------
From: Frank J. Wancho <wancho@brl>
To: tcp-ip at Brl
Subject: TCP/IP from scratch
Please forgive the following naive message, but I need to ask some basic
questions which don't appear to be addressed anywhere.
I have a couple of potential hosts in our lab of which no similar
machine already exists on the net. Both are SEL 32/xx series machines
which are capable of running a HASP background program (if that's worth
anything).
After reading through all of the existing documentation on TCP/IP
conversions for *existing* ARPANET hosts, it finally dawned on me that
those hosts still have the advantage of being able to use the existing
physical interface (1822), and here I thought that maybe "all" I had to
do was find some existing code (preferably in FORTRAN), and I'd have
those machines on the air. Not so simple...
Given that our SELs are able to run a HASP protocol, would the fact that
the C/30's optionally support HDLC be related? - I need some
enlightenment here - the point of mentioning HASP is that we have the
hardware to support a bisync interface at some arbitrarily high speed.
Of course, if this is not pertinent, then the rest of this message can
be ignored...
Second, and last question: is there a version of TCP/IP already written
under government sponsorship in any FORTRAN-77/78 dialect? Why FORTRAN?
Well, ours is a real-time FORTRAN designed to handle high-speed I/O on a
fully interrupt-driven architecture with separate I/O processors for
each major subsystem with "asynchronous" (no-wait) I/O capability. That
means we can put a program to sleep while it waits for I/O to complete
with an interrupt to wake it up. (All of this, I'm sure, is not a new
capability to most of you. All it means to me is that perhaps we can
run this mini as a direct host with minimum loading on the system and
our users.)
Any and all help encouraged!
--Frank
END OF TCP-IP DIGEST
********************
